GPT-Neo_DnD_Control / README.md
PatrickTyBrown's picture
update model card README.md
93953bf
|
raw
history blame
1.2 kB
metadata
tags:
  - generated_from_trainer
model-index:
  - name: GPT-Neo_DnD_Control
    results: []

GPT-Neo_DnD_Control

This model is a fine-tuned version of PatrickTyBrown/GPT-Neo_DnD on the None dataset. It achieves the following results on the evaluation set:

  • eval_loss: 2.6518
  • eval_runtime: 141.422
  • eval_samples_per_second: 6.527
  • eval_steps_per_second: 3.267
  • epoch: 3.9
  • step: 36000

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 1.5e-05
  • train_batch_size: 2
  • eval_batch_size: 2
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 5

Framework versions

  • Transformers 4.21.3
  • Pytorch 1.12.1+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1